neural-network

Testing neural network in C++

What is a Neural network

A Neural Network (ANN) is a paradigm inspired by the way biological systems process information, such as animal brains, working by using loads of interconnected neurons forming a giant network. The key element of this paradigm, like the brain it is imitating, is the novel structure of the information processing system - composed of many interconnected 'neurons' working in unison to solve specific problems. Computer based (Artificial) Neural Networks, like the brain, learn by example. A NN is specialised for a specific application, such as image processing or data classification. The brain learns by adjusting synaptic connexions between the neurons, this is also how artificial neural networks work.

About this ANN

This ANN is implemented in C++ using just standard libraries and no NN libraries. The two main structures of this program consist of the class neuron and the class network. In these objects we have the necessary variables and functions in order to form most of our network.

My Tests

So far I have used this back prop neural net to identify solutions to digital circuits, as the network will learn what an and gate and or gate etc will do, plus also for letter recognition of multiple fonts. In future I plan to use it for more applications and you should be able to do so as well with some minor tweaking.

Download and Setup

Download the files with

git clone git@github.com:dougbrion/back-propagating-neural-network.git

Enter the directory and make using

make bin/test

Also if you want make other binarys for changing data format and also creating the data.

make bin/change_char
make bin/creation

Use

Network is currently setup for basic training using ones and zeros. We can run the program using the following.

bin/test data.txt

Notice the training data is input as an argument.